Oxidized low-density lipoprotein, iron stores, and haptoglobin polymorphism

Atherosclerosis 176 (2004) 189–195

Oxidized low-density lipoprotein, iron stores, and haptoglobin polymorphism Annelies Brouwers a , Michel Langlois a,b,∗ , Joris Delanghe b , Johan Billiet a , Marc De Buyzere c , Raf Vercaemst a , Ernst Rietzschel c , Dirk Bernard a , Victor Blaton a a

Department of Clinical Chemistry and Haematology, AZ St-Jan AV Hospital, Ruddershove 10, B-8000 Bruges, Belgium b Department of Clinical Chemistry, Ghent University Hospital, De Pintelaan 185, B-9000 Ghent, Belgium c Department of Cardiovascular Diseases, Ghent University Hospital, De Pintelaan 185, B-9000 Ghent, Belgium Received 26 February 2004; accepted 12 May 2004 Available online 1 July 2004

Abstract Background: In vitro experimental studies demonstrated that iron promotes free radical-induced low-density lipoprotein (LDL) oxidation. Objective: To test the hypothesis that circulating oxidized LDL (oxLDL) levels might be associated with body iron stores (serum ferritin) and iron-related genetic markers (hemochromatosis gene C282Y mutation, haptoglobin polymorphism). Methods: We investigated 381 (176 males, 205 females, age 45 ± 6 years) healthy Caucasians. Serum oxLDL, assayed by a mAb-4E6-based enzyme-linked immunosorbent assay (ELISA), was expressed as oxLDL/LDL ratio to adjust for serum LDL-cholesterol concentration. Hemochromatosis gene C282Y mutation analysis was performed by a Taqman® -based polymerase chain reaction (PCR) assay. Haptoglobin (Hp) phenotypes (Hp 1-1, Hp 2-1, Hp 2-2) were determined by starch gel electrophoresis. Results: In stepwise multivariate regression analysis, gender (P < 0.0001), current smoking (P < 0.0001), HDL-cholesterol (P = 0.0001), ferritin (P = 0.0051), body mass index (BMI) (P = 0.0063), and Hp phenotype (P = 0.0331) independently predicted oxLDL/LDL ratio in the total group. In men, smoking (P < 0.0001), ferritin (P = 0.0052), Hp phenotype (P = 0.0063), and HDL-cholesterol (P = 0.0127) were independent determinants of oxLDL/LDL ratio. In women, only body mass index (P < 0.0001), HDL-cholesterol (P = 0.0005), and smoking (P = 0.0025) were significantly associated with oxLDL/LDL ratio. The C282Y mutation (wild-type versus C282Y heterozygotes) was not associated with oxLDL/LDL ratio in both sexes. Conclusion: Serum ferritin concentration and Hp polymorphism are independently associated with circulating oxLDL levels in males. © 2004 Elsevier Ireland Ltd. All rights reserved. Keywords: Oxidized LDL; Iron; Ferritin; Haptoglobin

1. Introduction Oxidized low-density lipoprotein (oxLDL) plays a key role in the initiation and progression of vascular atherosclerotic lesions [1]. Experimental studies have shown that oxidatively modified LDL particles are internalized in monocyte-derived macrophages through specific scavenger receptors, an event that leads to the formation and accumulation of lipid-loaded foam cells in the arterial wall [1]. Oxidation of polyunsaturated fatty acids by redox-active metals (e.g., iron) and reactive oxygen species is considered to be the initiating step in the modification of LDL [2]. Pro-oxidative forms of iron (Fe2+ ) and haem, derived ∗

Corresponding author. Tel.: +32 50 452640; fax: +32 50 452619. E-mail address: [email protected] (M. Langlois).

from haemoglobin (Hb), are able to generate agressive hydroxyl radicals in the presence of H2 O2 (Fenton reaction) which can initiate lipid peroxidation by hydrogen abstraction from the fatty acids [2,3]. This causes the formation of reactive lipid decomposition products such as malondialdehyde (MDA) which can react with lysine residues of LDL-associated apolipoprotein B-100, resulting in oxLDL that is recognized by the macrophage scavenger receptor [2,3]. In iron-containing culture media, LDL oxidation is further stimulated by various cell types, particularly monocyte macrophages and endothelial cells, which are known to produce superoxide radicals and H2 O2 [4]. These experimental data suggest that iron metabolism plays a role in lipid peroxidation in vivo. Clinically relevant iron overload was shown to increase lipid oxidation products in patients with ␤-thalassemia and hereditary hemochro-

0021-9150/$ – see front matter © 2004 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.atherosclerosis.2004.05.005

190

A. Brouwers et al. / Atherosclerosis 176 (2004) 189–195

matosis [5,6]. In men, reduction of body iron stores by phlebotomy significantly decreased susceptibility of LDL to ex vivo oxidation [7]. Epidemiologic studies, however, provide conflicting data about the role of iron metabolism in atherosclerotic vascular disease. Some but not all studies showed that serum ferritin, which reflects the level of total body iron stores, is related to coronary and carotid atherosclerosis [8–11]. Studies in subjects heterozygous for the cysteine-to-tyrosine mutation at amino acid position 282 (C282Y) within the hemochromatosis-associated gene HFE do not provide consistent evidence for an association between the mutation and the development of atherosclerosis [10,11]. Little is known about the influence of asymptomatic C282Y heterozygosity, if any, on in vivo lipid peroxidation. In recent years, other genes involved in iron turnover have been studied extensively. Among them, the genetic polymorphism of haptoglobin (Hp) has been proposed as an independent risk factor for atherosclerotic vascular disease [12–14]. Hp is a Hb-binding plasma protein mediating haem iron recycling following hemolysis [15]. It plays an important antioxidant role by forming a stable Hp–Hb complex and thereby preventing Hb-induced oxidative tissue damage [16]. In humans, three structurally different phenotypes are known (Hp 1-1, Hp 2-1, Hp 2-2) which result from the expression of two different alleles (Hp1, Hp2) of the Hp gene located on chromosome 16q22 [15]. The Hp 2-2 phenotype has less efficient antioxidant properties [17] and is associated with delocalized storage of Hb-derived iron in monocyte macrophages [18]. From these in vitro and experimental data, we hypothesized that circulating levels of oxLDL might be affected by iron stores in vivo. In this study, we investigated serum oxLDL concentration in healthy individuals and its relationship with body iron stores (serum ferritin) and iron-related genes (HFE, Hp).

Venous blood was sampled in empty vacuum tubes and EDTA-tubes (Terumo, Haasrode, Belgium) after overnight fasting. Serum and EDTA-plasma were obtained by centrifugation (1000 × g, 10 min) and kept frozen at −20 ◦ C until assayed. For HFE typing, blood was collected in a separate EDTA-tube that was also stored at −20 ◦ C. 2.2. Serum oxLDL concentration Serum oxLDL concentration was measured by an enzyme-linked immunosorbent assay (ELISA) based on a murine monoclonal antibody, mAb-4E6 [19,20], specific for a neo-epitope in the aldehyde-substituted lysine residues of the apolipoprotein B-100 moiety of oxLDL (Mercodia, Uppsala, Sweden). The bound oxLDL was detected with a peroxidase-conjugated anti-apolipoprotein B antibody and a colorimetric reaction with 3,3 ,5,5 -tetramethylbenzidine that was read at 450 nm. 2.3. Plasma MDA concentration Plasma MDA was assayed using a high performance liquid chromatography (HPLC) method based on the classic thiobarbituric acid (TBA) reaction [21]. An aliquot of 200 ␮l plasma was diluted with 750 ␮l H3 PO4 , 0.44 mol/l and mixed with 350 ␮l TBA, 42 mmol/l. After heating at 100 ◦ C for one hour, an aliquot of 20 ␮l was injected into the HPLC system (Merck LaChrom, Darmstadt, Germany). The TBA– MDA adduct was separated on a reversed-phase column (NOVA-pak C18 3.9 × 150 mm, Waters, Milford, MA) and monitored by fluorescence detection (λex = 515 nm, λem = 543 nm). The column was isocratically eluted at 1 ml/min with CH3 OH/0.6% KH2 PO4 pH 6.0 (30/70, v/v). The method was calibrated using 1,1,3,3-tetraethoxypropane as standard. 2.4. Other biochemical parameters

2. Materials and methods 2.1. Subjects and blood sampling This study included 381 (176 males and 205 females) Caucasian individuals from the region Erpe-mere and Nieuwerkerken (Belgium). The subjects participated in a primary prevention screening program (Asklepios study) in the age category of 35–55 years organized by local primary physicians. They were all free of overt cardiovascular diseases, type I diabetes mellitus, and all other life-threatening diseases. None of the subjects was anaemic or had excessive alcohol intake. At the time of blood sampling, subjects had to be free of acute infectious disease or other conditions possibly associated with an acute phase reaction. None of the female participants was pregnant. All subjects gave informed consent to participate in the community screening program and to undergo relevant genetic screenings concerning the underlying mechanisms of atherosclerosis.

Serum concentrations of total cholesterol, HDL-cholesterol, and triglycerides were determined by commercially available colorimetric-enzymatic methods on a Modular P800 analyser (Roche, Mannheim, Germany). Serum LDL-cholesterol concentrations were calculated using the Friedewald formula. Serum C-reactive protein (hs-CRP) concentrations were measured by particle-enhanced immunoturbidimetry (Integra 400, Roche). Serum ferritin was assayed by fixed-time immunonephelometry on a BN II nephelometer (Dade Behring, Marburg, Germany). 2.5. HFE C282Y mutation analysis The G-to-A transition at nucleotide 845 of the HFE gene, resulting in a substitution of tyrosine for cysteine at codon 282, was analysed by a TaqMan® -based polymerase chain reaction assay. DNA was prepared from 200 ␮l blood using the QiaAmp DNA Blood Mini Kit (Qiagen, Hilden, Ger-

A. Brouwers et al. / Atherosclerosis 176 (2004) 189–195

many). A volume of 2.5 ␮l DNA was amplified in a 25 ␮l reaction mixture containing TaqMan® Universal PCR Master Mix (Applied Biosystems, Foster City, CA), 25 pmol of each primer (G845Af and G845Ar) and 5 pmol of each probe (G845A wild-type and mutant probe). G845Af is 5 -GCT GGA TAA CCT TGG CTG TAC-3 , G845Ar is 5 -AGC TCC TGG CTC TCA TCA GT-3 , G845A wild-type probe is 5 -FAM-CAC CTG GCA CGT AT-MGB-3 , and G845A mutant probe is 5 -VIC-CAC CTG GTA CGT ATA T-MGB-3 . The universal cycling conditions of the ABI Prism® 7700 Sequence Detection Systems (Applied Biosystems) were used and the fluorescence was analysed using end-point measurement. 2.6. Determination of Hp phenotype Hp phenotyping was performed by means of starch gel electrophoresis of Hb-supplemented serum [22]. Briefly, human Hb was prepared by washing EDTA-blood three times with 0.15 mol/l NaCl, disrupting the erythrocytes with distilled water, and centrifugation (5000 × g, 20 min) to remove nuclei and ghosts. Fifty microliters of Hb (45 g/l) were then added to 500 ␮l serum. Starch gel was prepared using 11.5% hydrolysed starch in 0.1 mol/l Tris–citrate buffer, pH 8.86. Electrophoresis was performed during 1 h at 200 V in a 0.3 mol/l borate buffer, pH 8.4. Afterwards, the gel was cut horizontally along its length. The free Hb and the Hp–Hb complex bands were visualised by staining the gel slices with metal-enhanced peroxidase reagents (Pierce Corp., Rockford, USA). The Hp phenotype was determined from the relative migration position of slow and fast migrating Hp–Hb bands [15]. 2.7. Statistics Data are presented as mean ± S.D. or median (interquartile range) where appropriate. Circulating oxLDL was expressed as oxLDL/LDL ratio (U/mmol) to adjust for serum LDL-cholesterol concentration. Differences were evaluated using the Wilcoxon test for comparison between two subgroups and the Kruskal–Wallis test for >2 subgroups. Correlations between parameters were examined using Spearman rank analysis. Age, gender, current smoking, body mass index (BMI), HDL-cholesterol, triglycerides, ferritin, hs-CRP, C282Y mutation (wild-type versus C282Y heterozygotes), and Hp phenotype were included in stepwise multivariate regression analyses with oxLDL/LDL ratio as dependent variable. Statistical analysis was carried out using MEDCALC software (Mariakerke, Belgium), version 6.12. Statistical significance was considered at the level of P < 0.05.

3. Results Serum oxLDL concentration was 79.4 ± 31.1 U/l (median 76.4 U/l) in the total study population and was not signif-

191

Table 1 Age, smoking, BMI, and biochemical parameters according to gender in the study population Parameter

Males (n = 176)

Females (n = 205)

Age (years) Current smokers (%) BMI (kg/m2 ) Total cholesterol (mmol/l) HDL-cholesterol (mmol/l) LDL-cholesterol (mmol/l) Triglycerides (mmol/l) OxLDL/LDL ratio (U/mmol) MDA (␮mol/l) hs-CRP (mg/l) Ferritin (␮g/l)

44.9 ± 5.9 18 26.0 ± 4.1 5.24 ± 0.86 1.34 ± 0.36 3.24 ± 0.75 1.24 (0.80–1.81) 26.0 ± 9.4 0.69 ± 0.26 1.58 (0.84–3.41) 135 (108–203)

44.5 ± 6.0 23 27.7 ± 3.9∗∗ 5.19 ± 0.78 1.62 ± 0.44∗∗ 3.10 ± 0.72∗ 1.07 (0.74–1.59)∗ 25.0 ± 9.6 0.63 ± 0.24 1.55 (0.78–3.25) 57 (35–84)∗∗∗

Data are presented as mean ± S.D. or median (interquartile range) where appropriate. BMI, body mass index; HDL, high-density lipoprotein; LDL, low-density lipoprotein; oxLDL, oxidized low-density lipoprotein; MDA, malondialdehyde; hs-CRP, C-reactive protein. ∗ P < 0.05. ∗∗ P < 0.005. ∗∗∗ P < 0.0001, comparing males vs. females (Wilcoxon test).

icantly different between men and women. Serum oxLDL concentrations correlated positively with total cholesterol (r = 0.219, P < 0.0001) and LDL-cholesterol (r = 0.307, P < 0.0001) levels. For this reason, serum oxLDL was normalized for LDL-cholesterol concentration in all further analyses. Table 1 summarizes oxLDL/LDL ratio (U/mmol) as well as age, gender, BMI, and other biochemical data in both sexes. Serum oxLDL/LDL ratio was higher in smokers (34.1 ± 11.3 U/mmol) than in non-smokers (24.2 ± 9.0 U/mmol, P < 0.0001). In the total group, a negative correlation was observed between serum oxLDL/LDL ratio and HDL-cholesterol concentration (r = −0.313, P < 0.0001). Overall, serum oxLDL/LDL ratio correlated positively with plasma MDA (r = 0.342, P = 0.0009). Serum ferritin concentrations were significantly higher in men than in women (P < 0.0001) and correlated weakly with hs-CRP concentration (r = 0.165, P = 0.042). Serum hs-CRP concentrations were <10 mg/l in all individuals, and also correlated with BMI (r = 0.152, P = 0.034) and HDL-cholesterol (r = −0.207, P = 0.004), but not with oxLDL/LDL ratio or plasma MDA. The prevalence of the HFE C282Y mutation and Hp phenotypes among the study participants as well as the calculated relative C282Y allele and Hp1 allele frequencies are shown in Table 2. None of the subjects was anhaptoglobinaemic. A stepwise multivariate regression analysis was performed with oxLDL/LDL ratio as dependent variable in the total study group (Table 3A). In this model, gender (P < 0.0001), smoking (P < 0.0001), HDL-cholesterol (P < 0.0001), ferritin (P = 0.0051), BMI (P = 0.0063), and Hp phenotype (P = 0.0331) emerged as significant predictors of oxLDL/LDL ratio while age, triglycerides, hs-CRP, and the

192

A. Brouwers et al. / Atherosclerosis 176 (2004) 189–195

Table 2 Distributions of the HFE C282Y mutation and Hp phenotypes Gene

Table 4 Serum oxLDL/LDL ratio according to quartiles of serum ferritin in men (n = 176)

Frequency (%) Total

Males

Females

HFE C282Y mutation wt/wt C282Y/wt C282Y/C282Y C282Y allele frequency

84.0 15.2 0.8 0.08

87.5 12.5 0.0 0.06

81.0 17.5 1.5 0.10

Hp phenotype Hp 1-1 Hp 2-1 Hp 2-2 Hp1 allele frequency

18.6 47.5 33.9 0.42

20.4 47.2 32.4 0.44

17.1 47.8 35.1 0.41

Quartiles of serum ferritin (␮g/l)

OxLDL/LDL ratio (U/mmol)

<108 108–135 136–203 >203

22.8 25.0 26.1 31.8

± ± ± ±

8.6∗∗ 10.9∗ 9.1∗ 8.5

Data are presented as mean ± S.D. ∗ P < 0.01, compared to the highest quartile (Wilcoxon test). ∗∗ P < 0.005, compared to the highest quartile (Wilcoxon test). Table 5 Serum oxLDL/LDL ratio and ferritin concentration according to Hp phenotype in men (n = 176)

Hp, haptoglobin; HFE, hemochromatosis-associated gene; wt, wild-type.

C282Y mutation (wild-type versus C282Y heterozygotes) were not significantly associated. Stepwise multivariate regression analysis for serum oxLDL/LDL ratio in the male subgroup (Table 3B) showed that smoking (P < 0.0001), ferritin (P = 0.0052), Hp phenotype (P = 0.0063), and HDL-cholesterol (P = 0.0127) emerged as independent determinants of oxLDL/LDL ratio in men. In contrast, this association could not be observed in women (Table 3C); in this subgroup BMI (P < 0.0001) was significantly associated with oxLDL/LDL ratio as well as HDL-cholesterol (P = 0.0005) and smoking (P = 0.0025). Table 3 Stepwise multivariate regression analysis for serum oxLDL/LDL ratio (U/mmol) in the total group (A), males (B), and females (C) β (S.E.)

t-Statistic

Significance

Totala

(A) Gender (male vs. Female) Smoking (yes vs. no) HDL-cholesterol (mmol/l) Ferritin (␮g/l) BMI (kg/m2 ) Hp phenotype

7.88 10.96 −8.03 0.03 0.44 2.69

(1.79) (1.59) (1.65) (0.01) (0.16) (0.91)

4.38 6.85 −4.86 2.83 2.75 2.96

<0.0001 <0.0001 0.0001 0.0051 0.0063 0.0331

(B) Menb Smoking (yes vs. no) Ferritin (␮g/l) Hp phenotype HDL-cholesterol (mmol/l)

16.46 0.04 4.02 −5.20

(1.80) (0.01) (0.98) (2.05)

9.16 3.58 4.09 −2.53

<0.0001 0.0052 0.0063 0.0127

(C) Womenc BMI (kg/m2 ) HDL-cholesterol (mmol/l) Smoking (yes vs. no)

0.91 (0.25) −9.79 (2.37) 5.67 (2.51)

3.59 −4.13 2.26

<0.0001 0.0005 0.0025

a F-ratio = 22.62; P < 0.001 for the model. Variables that were not significantly associated: age, triglycerides, hs-CRP, HFE C282Y mutation (wild-type vs. C282Y heterozygotes). b F-ratio = 26.64; P < 0.001 for the model. Variables that were not significantly associated: age, BMI, triglycerides, hs-CRP, HFE C282Y mutation (wild-type vs. C282Y heterozygotes). c F-ratio = 10.03; P < 0.001 for the model. Variables that were not significantly associated: age, triglycerides, hs-CRP, ferritin, Hp phenotype, HFE C282Y mutation (wild-type vs. C282Y heterozygotes).

Hp phenotype

OxLDL/LDL (U/mmol)

Ferritin (␮g/l)

Hp 1-1 Hp 2-1 Hp 2-2 Kruskal–Wallis test

19.8 ± 9.1 25.4 ± 11.5 28.5 ± 10.1 P = 0.006

129 (103–175) 135 (104–234) 143 (114–230) NS

Data are mean ± S.D. and median (interquartile range). Hp, haptoglobin; oxLDL, oxidized low-density lipoprotein; NS, not significant.

A positive correlation was observed between oxLDL/LDL ratio and BMI in women (r = 0.319, P < 0.0001). In men, serum oxLDL/LDL ratio correlated positively with ferritin concentration (r = 0.397, P < 0.0001). Men with ferritin concentration in the upper quartile (>203 ␮g/l) had a significantly higher oxLDL/LDL ratio compared to those with ferritin concentration ≤ 203 ␮g/l (P = 0.003). Table 4 shows oxLDL/LDL ratio according to quartiles of serum ferritin in the male subgroup. In men, serum ferritin also correlated positively with plasma MDA (r = 0.315, P = 0.035). In contrast, these relationships could not be observed in the female subgroup. Serum oxLDL/LDL ratio, plasma MDA, and serum ferritin concentration were not significantly different between HFE wild-types and C282Y heterozygotes in both male and female subgroups (data not shown). Analysis with respect to Hp polymorphism, however, showed that men with a Hp 2-2 phenotype had significantly higher oxLDL/LDL ratios compared to those carrying Hp 2-1 or Hp 1-1 (P = 0.006) whereas serum ferritin concentrations were not significantly different (Table 5). This Hp phenotype-dependent variation of oxLDL/LDL ratio could not be observed in the female subgroup. Plasma MDA concentrations were not different between Hp phenotypes in both sexes (data not shown).

4. Discussion Circulating oxLDL is strongly associated with atherosclerotic vascular disease [19,20]. Oxidation of LDL occurs under the influence of reactive oxygen species and the catalytic activity of transition metals such as iron, initiating the alteration of its apolipoprotein B into a ligand for the scavenger

A. Brouwers et al. / Atherosclerosis 176 (2004) 189–195

receptor of macrophages that temporarily reside in the arterial intima [1,2]. The oxLDL uptake results in the conversion of macrophages into foam cells, starting the formation of the early atheromatous lesion [1]. In this study, we used a recently developed sandwich ELISA for the direct measurement of serum oxLDL concentration. The method is based on the same specific murine monoclonal 4E6 antibody as Holvoet et al. in their assays [19,20], but the latter used a competitive instead of a sandwich ELISA. In our study, levels were normalized using oxLDL/LDL ratios to adjust for serum LDL-cholesterol concentration. In our study group of healthy individuals, serum oxLDL/LDL ratio showed a positive correlation with plasma levels of MDA, which is a stable oxidative degradation product of polyunsaturated fatty acids and is widely recognized as a marker for lipid peroxidation [2]. The method for MDA determination was based on the classic TBA reaction, which is not very specific for MDA, but also reacts with various sugars and amino acids, generally called thiobarbituric acid-reactive substances (TBARS) [2]. In this study, HPLC was used to avoid this lack of specificity as it allows analytical separation of the TBA–MDA complex from interfering substances [21]. Determinants of oxLDL/LDL ratio assessed by stepwise multivariate analysis in the total population were gender, BMI, current smoking, HDL-cholesterol, ferritin, and Hp phenotype. Among all individuals, we found that oxLDL/LDL ratio was higher in smokers and correlated negatively with serum concentration of HDL-cholesterol, which is generally believed to be a cardioprotective lipid particle. Normal HDL has been shown to inhibit LDL oxidation [1]. In men, we could observe a positive correlation between oxLDL/LDL ratio and serum ferritin levels. Men with ferritin concentrations >203 ␮g/l showed a high oxLDL/LDL ratio, while differences were less pronounced within the lower range of ferritin concentration. Serum ferritin also correlated positively with plasma MDA in men. These findings are consistent with data from previous studies that reported a correlation of serum ferritin with TBARS in patients with hereditary hemochromatosis [6] and with lipid hydroperoxides in thalassemia [5]. A protective effect of iron depletion was suggested by the fact that reduction of body iron stores by phlebotomy increased the resistance to oxidation of serum lipoproteins ex vivo [7]. Other studies, however, failed to demonstrate an association between serum ferritin and in vitro measures of LDL oxidation [10,23]. The role of body iron stores in atherosclerosis remains controversial. In vitro experimental studies suggest that iron and haem compounds can oxidize LDL into a form that is recognized by the macrophage scavenger receptor [2,3]. The hydroxyl radical, formed by the interaction between iron and H2 O2 , is a highly reactive oxidant initiating lipid peroxidation through abstraction of a hydrogen atom from polyunsaturated fatty acids [2]. In vivo, the vast ma-

193

jority of iron does not exist in free form but is contained, mainly intracellularly, within Hb (70%) and ferritin (20%) [24]. Increased ferritin gene expression has been observed in macrophages and endothelial cells in atherosclerotic lesions [25]. Cell culture studies demonstrated that iron is exocytosed from iron- or Hb-loaded macrophages, and such released iron may promote LDL oxidation and uptake by macrophages in the arterial wall [26]. Some epidemiological studies provided evidence for the so-called “iron hypothesis” reporting an association between iron stores and the progression of atherosclerotic vascular disease in men, but many other studies did not [8–10]. This lack of consistency is probably explained by the large variability in biochemical estimates of iron status, by ethnic diversity, by the diversity in study outcomes, and by the potential influence of non-genetic factors that increase or decrease iron stores. Ferritin is a high-capacity, multimeric protein that serves as the body’s storage site for iron mainly in hepatocytes and macrophages, and because serum ferritin concentrations are directly proportional to intracellular ferritin content, it is considered to be the best clinical indicator of body iron stores [24]. However, it is important to rule out non-iron-related factors and underlying conditions that can affect serum ferritin concentration such as alcohol intake and inflammation [24]. Among our participants, however, excessive alcohol intake was ruled out and hs-CRP concentrations were <10 mg/l in all individuals. In our study, we could not observe an association between heterozygosity for the hemochromatosis gene C282Y mutation and oxLDL/LDL ratio or plasma MDA in males nor in females. This suggests that lipid oxidation, at least in non-hemochromatotic subjects, is not influenced by C282Y heterozygosity, a proposed risk factor for atherosclerotic vascular disease [10,11]. The Hp polymorphism has also been proposed as one of the candidate genes in the multigenic model of atherogenesis [15]. Men with a Hp 2-2 phenotype are more at risk for developing premature coronary artery disease [12], peripheral vascular disease [13], and vascular diabetic complications [14]. We found that oxLDL/LDL ratio was significantly higher among men with a Hp 2-2 phenotype compared to those carrying the Hp 1-1 type, whereas the heterozygotes (Hp 2-1) showed intermediate levels. In contrast, plasma MDA concentrations were not different between Hp phenotypes. Previous reports have suggested a link between Hp polymorphism and iron-driven oxidative stress. In vivo, Hp has an antioxidant function by binding free Hb that continuously leaks from erythrocytes [15]. Under conditions that increase erythrocyte fragility, at sites of turbulent blood flow or at locations of microhemorrages, this scavenging system may be saturated [15]. Recent studies suggest that oxidative interactions of Hb may be more important than iron in the oxidative modification of LDL [16]. Haem, released from Hb, binds to the lipoprotein surface monolayer with high affinity and catalyzes the oxidation of LDL core lipids in the presence of H2 O2 (Fenton chemistry) [3]. The Hp 2-2 phe-

194

A. Brouwers et al. / Atherosclerosis 176 (2004) 189–195

notype is a less efficient antioxidant because it has a lower Hb binding capacity, which is reflected by low vitamin C stability in plasma [27] and higher susceptibility of LDL to in vitro oxidation (lag phase duration) than Hp 1-1 during antioxidative treatment [28]. The Hp phenotypes also differ in their ability to remove free Hb from extravascular compartments, particularly after endothelial injury [15]. Indeed, Hp 2-2 polymers (>200 kDa) are characterized by a limited penetration into the extravascular space across the endothelial cell barrier and thus a less efficient Hb clearance than Hp 1-1 dimers (86 kDa) [15,17]. Furthermore, complexes of Hb and Hp 2-2 exhibit higher affinity for the Hb-scavenger receptor CD163 on monocyte macrophages [29], resulting in haem iron retention and increased ferritin expression in macrophages [18]. Such macrophages may, in the microenvironment of the arterial intima, expose LDL to a locally high concentration of pro-oxidant iron forms and reactive oxygen species [26]. In our study, Hp phenotype and serum ferritin were independent predictors of oxLDL/LDL ratio in men. This suggests that the influence of Hp polymorphism on circulating oxLDL could be mediated through other pathways than iron turnover. Indeed, Hp is a multifunctional protein and thus other functional properties of the Hp phenotypes, e.g. immunomodulatory effects [15], might play a role in the sequence of events leading to LDL oxidation. In women, we could not observe a significant association of oxLDL/LDL ratio with serum ferritin concentration and Hp phenotype, although the mean oxLDL/LDL ratio was not different between men and women despite a comparable number of smokers and higher serum HDL-cholesterol concentrations in the latter subgroup. Multiple regression analysis revealed that BMI strongly contributed to oxLDL/LDL ratio in women. BMI is one of the components of the metabolic syndrome, which has been associated with worse antioxidant/oxidant balance [30]. We found that BMI correlated positively with oxLDL/LDL ratio in women whereas serum triglycerides, another component of the metabolic syndrome, were not related. In summary, we could demonstrate that higher serum ferritin concentrations and the Hp 2-2 phenotype are independently associated with higher circulating oxLDL levels in men. Our results therefore might support the role of iron-driven oxidative stress and Hp polymorphism in lipid peroxidation in vivo.

Acknowledgements We thank Martin Ballegeer, Frida Brusselmans, Lutgard Claeys, Els Mahieu, Femke Van Hoeke, Ellen Van Neder, and Roger Verhoeven for their expert technical assistance. This study is part of the postdoctoral research (FWO-Flanders, Belgium) of Michel Langlois. Ernst Rietzschel is primary investigator of the Asklepios Study (G042703) organized by primary physicians from Erpe-Mere and Nieuwerkerken.

References [1] Mertens A, Holvoet P. Oxidized LDL and HDL: antagonists in atherothrombosis. FASEB J 2001;15:2073–84. [2] Jialal I, Devaraj S. Low-density lipoprotein oxidation, antioxidants and atherosclerosis: a clinical biochemistry perspective. Clin Chem 1996;42:498–506. [3] Camejo G, Halberg C, Manschik-Lundin A, et al. Hemin binding and oxidation of lipoproteins in serum: mechanisms and effect on the interaction of LDL with human macrophages. J Lipid Res 1998;39:755–66. [4] Chisolm GM, Hazen SL, Fox PL, Cathcart MK. The oxidation of lipoproteins by monocytes-macrophages. Biochemical and biological mechanisms. J Biol Chem 1999;274:25959–62. [5] Livrea MA, Tesoriere L, Maggio A, et al. Oxidative modification of low-density lipoprotein and atherogenic risk in ␤-thalassemia. Blood 1998;92:3936–42. [6] Gaenzer H, Marschang P, Sturm W, et al. Association between increased iron stores and impaired endothelial function in patients with hereditary hemochromatosis. J Am Coll Cardiol 2002;40:2189– 94. [7] Salonen JT, Korpela H, Nyyssonen K, et al. Lowering of body iron stores by blood letting and oxidation resistance of serum lipoproteins: a randomized cross-over trial in male smokers. J Intern Med 1995;237:161–8. [8] Meyers DG. The iron hypothesis—does iron cause atherosclerosis? Clin Cardiol 1996;19:925–9. [9] Wolff B, Völzke H, Lüdemann J, et al. Association between high serum ferritin levels and carotid atherosclerosis in the study of health in Pomerania (SHIP). Stroke 2004;35:453–7. [10] Bozzini C, Girelli D, Tinazzi E, et al. Biochemical and genetic markers of iron status and the risk of coronary artery disease: an angiography-based study. Clin Chem 2002;48:622–8. [11] Rossi E, McQuillan BM, Hung J, et al. Serum ferritin and C282Y mutation of the hemochromatosis gene as predictors of asymptomatic carotid atherosclerosis in a community population. Stroke 2000;31:3015–20. [12] Delanghe JR, Cambier B, Langlois MR, et al. Haptoglobin polymorphism, a genetic risk factor in coronary artery bypass surgery. Atherosclerosis 1997;132:215–9. [13] Delanghe J, Langlois M, Duprez D, De Buyzere M, Clement D. Haptoglobin polymorphism and peripheral arterial occlusive disease. Atherosclerosis 1999;145:287–92. [14] Levy AP, Roguin A, Hochberg I, et al. Haptoglobin phenotype and vascular complications in patients with diabetes. N Engl J Med 2000;343:969–70. [15] Langlois MR, Delanghe JR. Biological and clinical significance of haptoglobin polymorphism in humans. Clin Chem 1996;42:1589– 600. [16] Miller YI, Altamentova SM, Shaklai N. Oxidation of low-density lipoprotein by hemoglobin stems from a heme-initiated globin radical: antioxidant role of haptoglobin. Biochemistry 1997;36:12189– 98. [17] Melamed-Frank M, Lache O, Enav BI, et al. Structure-function analysis of the antioxidant properties of haptoglobin. Blood 2001;98:3693– 8. [18] Langlois MR, Martin ME, Boelaert JR, et al. The haptoglobin 2-2 phenotype affects serum markers of iron status in healthy males. Clin Chem 2000;46:1619–25. [19] Holvoet P, Vanhaecke J, Janssens S, Van de Werf F, Collen D. Oxidized LDL and malondialdehyde-modified LDL in patients with acute coronary syndromes and stable coronary artery disease. Circulation 1998;98:1487–94. [20] Holvoet P, Mertens A, Verhamme P, et al. Circulating oxidized LDL is a useful marker for identifying patients with coronary artery disease. Arterioscler Thromb Vasc Biol 2001;21:844–8.

A. Brouwers et al. / Atherosclerosis 176 (2004) 189–195 [21] Wong SHY, Knight JA, Hopfer SM, et al. Lipoperoxides in plasma as measured by liquid-chromatographic separation of malondialdehyde–thiobarbituric acid adduct. Clin Chem 1987;33:214–20. [22] Smithies O. Zone electrophoresis in starch gels: group variations in the serum proteins of normal human adults. Biochem J 1955;61:629– 41. [23] Iribarren C, Sempos CT, Eckfeldt JH, Folsom AR. Lack of association between ferritin levels and measures of LDL oxidation: the ARIC study. Atherosclerosis risk in communities. Atherosclerosis 1998;139:189–95. [24] Baynes RD. Assessment of iron status. Clin Biochem 1996;29:209– 15. [25] Pang J-H, Jiang M-J, Chen Y-L, et al. Increased ferritin expression in atherosclerotic lesions. J Clin Invest 1996;97:2204–12.

195

[26] Yuan XM, Brunk UT, Olsson AG. Effects of iron- and haemoglobin-loaded human monocyte-derived macrophages on oxidation and uptake of LDL. Arterioscler Thromb Vasc Biol 1995;15:1345–51. [27] Langlois MR, Delanghe JR, De Buyzere ML, Bernard D, Ouyang J. Effect of haptoglobin on the metabolism of vitamin C. Am J Clin Nutr 1997;66:606–10. [28] Bernard D, Christophe A, Delanghe J, et al. The effect of supplementation with an antioxidant preparation on LDL-oxidation is determined by haptoglobin polymorphism. Redox Rep 2003;8:41–6. [29] Kristiansen M, Graversen JH, Jacobsen C. Identification of the haemoglobin scavenger receptor. Nature 2001;409:198–201. [30] Senti M, Tomas M, Fito M, et al. Antioxidant paraoxonase 1 activity in the metabolic syndrome. J Clin Endocrinol Metab 2003;88:5422– 6.